Redo Tiling

In the current code, tiling and bilerp sampling are strongly tied together. They can be separated by taking advantage of observation that translating a sample point into filter points in the bilerp stage the filter points will be at most 0.5 outside the tile. This allows simplified repositioning for the various tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0-> max. This allows bilerp to simply treat the filter points that fall off the tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:5566
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed: https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
Committed: https://skia.googlesource.com/skia/+/86a6c6d8bdb1055ee2d0689eeaa4a814402e69dd

[herb_g, 2016-07-08 21:30:54]

Description was changed from

==========
Redo Tiling

BUG=skia:
==========

to

==========
Redo Tiling

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
==========

[herb_g, 2016-07-11 21:53:53]

WIP: Part way through bilerpers.

[herb_g, 2016-07-12 22:11:33]

WIP: draft of slow rate done.

[herb_g, 2016-07-13 21:47:13]

WIP: fixed clamp tiling.

[herb_g, 2016-07-13 21:47:14]

Description was changed from

==========
Redo Tiling

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
==========

to

==========
Redo Tiling

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
==========

[herb_g, 2016-07-14 22:01:17]

WIP: linear_pipeline working.

[herb_g, 2016-07-14 22:01:17]

Description was changed from

==========
Redo Tiling

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
==========

to

==========
Redo Tiling

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
==========

[herb_g, 2016-07-15 21:20:52]

WIP: unit working.

[herb_g, 2016-07-18 21:58:12]

All the blitters rewrittern.

[herb_g, 2016-07-18 22:12:51]

WIP: not passing DM but all written.

[herb_g, 2016-07-19 19:10:10]

Sync.

[herb_g, 2016-07-19 19:11:35]

The CQ bit was checked by herb@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-07-19 19:11:38]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-19 19:11:39]

Your CL relies on deprecated CQ feature(s):
 * Specifying master names in CQ_INCLUDE_TRYBOTS part of description without
"master." prefix is deprecated:
     client.skia
   For more details, see http://crbug.com/617627.

[commit-bot: I haz the power, 2016-07-19 19:15:08]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-07-19 19:15:08]

Dry run: Try jobs failed on following builders:
  Build-Ubuntu-GCC-Arm7-Debug-Android-Trybot on master.client.skia.compile
(JOB_FAILED,
http://build.chromium.org/p/client.skia.compile/builders/Build-Ubuntu-GCC-Arm...)

[herb_g, 2016-07-20 18:14:44]

fix zero rate.

[herb_g, 2016-07-20 18:14:59]

The CQ bit was checked by herb@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-07-20 18:15:11]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-20 18:15:14]

Your CL relies on deprecated CQ feature(s):
 * Specifying master names in CQ_INCLUDE_TRYBOTS part of description without
"master." prefix is deprecated:
     client.skia
   For more details, see http://crbug.com/617627.

[commit-bot: I haz the power, 2016-07-20 18:16:44]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-07-20 18:16:44]

Dry run: Try jobs failed on following builders:
  skia_presubmit-Trybot on master.client.skia.fyi (JOB_FAILED,
http://build.chromium.org/p/client.skia.fyi/builders/skia_presubmit-Trybot/bu...)

[herb_g, 2016-07-20 18:50:17]

Remove code.

[herb_g, 2016-07-20 18:57:28]

sync

[herb_g, 2016-07-20 18:57:46]

The CQ bit was checked by herb@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-07-20 18:57:57]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-20 18:58:02]

Your CL relies on deprecated CQ feature(s):
 * Specifying master names in CQ_INCLUDE_TRYBOTS part of description without
"master." prefix is deprecated:
     client.skia
   For more details, see http://crbug.com/617627.

[commit-bot: I haz the power, 2016-07-20 19:29:07]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-07-20 19:29:08]

Dry run: This issue passed the CQ dry run.

[herb_g, 2016-07-20 19:33:23]

Description was changed from

==========
Redo Tiling

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
==========

to

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13065155  8276022  0.633442  top25desk_google_com_search_q_c.skp_1
4843451   3311734  0.683755  top25desk_wordpress.skp_1
6895223   5822791  0.844467  top25desk_youtube_com.skp_1
3773279   3379005  0.895509  top25desk_google_com__hl_en_q_b.skp_1
560676    542446   0.967486  top25desk_techcrunch_com.skp_1
8863283   8649477  0.975877  top25desk_answers_yahoo_com.skp_1
2899749   2885793  0.995187  top25desk_google_com_calendar_.skp_1
2120466   2114308  0.997096  top25desk_amazon_com.skp_1
2427159   2427831  1.00028   top25desk_ebay_com.skp_1
3219321   3220414  1.00034   top25desk_facebook.skp_1
4474967   4480748  1.00129   top25desk_espn.skp_1
828955    836929   1.00962   top25desk_blogger.skp_1
4139912   4195259  1.01337   top25desk_linkedin.skp_1
5530509   5636178  1.01911   top25desk_sports_yahoo_com_.skp_1
4147545   4232179  1.02041   top25desk_news_yahoo_com.skp_1
996510    1020702  1.02428   top25desk_games_yahoo_com.skp_1
4224438   4345881  1.02875   top25desk_mail_google_com_mail_.skp_1
2686938   2773712  1.03229   top25desk_plus_google_com_11003.skp_1
3780211   3919857  1.03694   top25desk_docs___1_open_documen.skp_1
7258538   7699635  1.06077   top25desk_pinterest.skp_1
5177836   5515963  1.0653    top25desk_weather_com.skp_1
8833804   9444707  1.06916   top25desk_twitter.skp_1
940329    1027695  1.09291   top25desk_wikipedia__1_tab_.skp_1
7208273   7948446  1.10268   top25desk_booking_com.skp_1


BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
==========

[herb_g, 2016-07-20 19:33:41]

herb@google.com changed reviewers:
+ mtklein@google.com

[herb_g, 2016-07-21 14:29:02]

Description was changed from

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13065155  8276022  0.633442  top25desk_google_com_search_q_c.skp_1
4843451   3311734  0.683755  top25desk_wordpress.skp_1
6895223   5822791  0.844467  top25desk_youtube_com.skp_1
3773279   3379005  0.895509  top25desk_google_com__hl_en_q_b.skp_1
560676    542446   0.967486  top25desk_techcrunch_com.skp_1
8863283   8649477  0.975877  top25desk_answers_yahoo_com.skp_1
2899749   2885793  0.995187  top25desk_google_com_calendar_.skp_1
2120466   2114308  0.997096  top25desk_amazon_com.skp_1
2427159   2427831  1.00028   top25desk_ebay_com.skp_1
3219321   3220414  1.00034   top25desk_facebook.skp_1
4474967   4480748  1.00129   top25desk_espn.skp_1
828955    836929   1.00962   top25desk_blogger.skp_1
4139912   4195259  1.01337   top25desk_linkedin.skp_1
5530509   5636178  1.01911   top25desk_sports_yahoo_com_.skp_1
4147545   4232179  1.02041   top25desk_news_yahoo_com.skp_1
996510    1020702  1.02428   top25desk_games_yahoo_com.skp_1
4224438   4345881  1.02875   top25desk_mail_google_com_mail_.skp_1
2686938   2773712  1.03229   top25desk_plus_google_com_11003.skp_1
3780211   3919857  1.03694   top25desk_docs___1_open_documen.skp_1
7258538   7699635  1.06077   top25desk_pinterest.skp_1
5177836   5515963  1.0653    top25desk_weather_com.skp_1
8833804   9444707  1.06916   top25desk_twitter.skp_1
940329    1027695  1.09291   top25desk_wikipedia__1_tab_.skp_1
7208273   7948446  1.10268   top25desk_booking_com.skp_1


BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
==========

to

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
==========

[herb_g, 2016-07-21 14:32:56]

Fix perf problem.

[herb_g, 2016-07-21 14:33:35]

The CQ bit was checked by herb@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-07-21 14:33:44]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-21 14:33:46]

Your CL relies on deprecated CQ feature(s):
 * Specifying master names in CQ_INCLUDE_TRYBOTS part of description without
"master." prefix is deprecated:
     client.skia
   For more details, see http://crbug.com/617627.

[commit-bot: I haz the power, 2016-07-21 15:01:39]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-07-21 15:01:40]

Dry run: This issue passed the CQ dry run.

[mtklein, 2016-07-21 19:19:25]

lgtm

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
File src/core/SkLinearBitmapPipeline_sample.h (right):

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:443: // From an edgType, the integer
value of a pixel vs, and the integer value of the extreme edge
edgeType

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:551: SkScalar absLength =
SkScalarAbs(length);
May be clearer to calculate dx here and write the cases in terms of dx, or
explain why we don't.

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:559: if
(sample_to_filter(span.startX()) == 1.0f) {
if (s_to_f(x) == 1 && s_to_f(y) == 1) {
   src_strategy(...)
} else {
   this->spanUnitRate(span);
}

?

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:574: if (absLength < 2.0f * (count -
1)) {
might as well lift this if into the containing if

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:745: filterX -= dx;
// remember, dx < 0

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:980: filterX -= dx;
// remember, dx < 0, so filterX is increasing.

[herb_g, 2016-07-21 19:48:16]

Address comments.

[herb_g, 2016-07-21 19:51:05]

The CQ bit was checked by herb@google.com

[herb_g, 2016-07-21 19:51:05]

The patchset sent to the CQ was uploaded after l-g-t-m from mtklein@google.com
Link to the patchset: https://codereview.chromium.org/2134893002/#ps260001
(title: "Address comments.")

[commit-bot: I haz the power, 2016-07-21 19:51:17]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-21 19:51:20]

Your CL relies on deprecated CQ feature(s):
 * Specifying master names in CQ_INCLUDE_TRYBOTS part of description without
"master." prefix is deprecated:
     client.skia
   For more details, see http://crbug.com/617627.

[herb_g, 2016-07-21 19:52:20]

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
File src/core/SkLinearBitmapPipeline_sample.h (right):

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:443: // From an edgType, the integer
value of a pixel vs, and the integer value of the extreme edge
On 2016/07/21 19:19:25, mtklein wrote:
> edgeType

Done.

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:551: SkScalar absLength =
SkScalarAbs(length);
On 2016/07/21 19:19:25, mtklein wrote:
> May be clearer to calculate dx here and write the cases in terms of dx, or
> explain why we don't.

Done.

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:559: if
(sample_to_filter(span.startX()) == 1.0f) {
On 2016/07/21 19:19:25, mtklein wrote:
> if (s_to_f(x) == 1 && s_to_f(y) == 1) {
>    src_strategy(...)
> } else {
>    this->spanUnitRate(span);
> }
> 
> ?

Done.

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:574: if (absLength < 2.0f * (count -
1)) {
On 2016/07/21 19:19:25, mtklein wrote:
> might as well lift this if into the containing if

Done.

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:745: filterX -= dx;
On 2016/07/21 19:19:25, mtklein wrote:
> // remember, dx < 0

Done.

https://codereview.chromium.org/2134893002/diff/240001/src/core/SkLinearBitma...
src/core/SkLinearBitmapPipeline_sample.h:980: filterX -= dx;
On 2016/07/21 19:19:25, mtklein wrote:
> // remember, dx < 0, so filterX is increasing.

Done.

[commit-bot: I haz the power, 2016-07-21 20:22:07]

Description was changed from

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
==========

to

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed:
https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
==========

[commit-bot: I haz the power, 2016-07-21 20:22:08]

Committed patchset #14 (id:260001) as
https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1

[jcgregorio, 2016-07-22 12:40:46]

A revert of this CL (patchset #14 id:260001) has been created in
https://codereview.chromium.org/2174793002/ by jcgregorio@google.com.

The reason for reverting is: Crashing on Win with:

Caught exception 3221225477 EXCEPTION_ACCESS_VIOLATION, was running:
	unit test  GrShape
	srgb gm  shadertext2
	srgb gm  shallow_gradient_conical
	srgb gm  shallow_gradient_sweep
	srgb gm  shallow_gradient_linear_nodither
step returned non-zero exit code: -1073741819


https://status.skia.org/?commit_label=author&filter=search&search_value=Test-...
.

[herb_g, 2016-07-22 15:42:35]

Description was changed from

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed:
https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
==========

to

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed:
https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
==========

[herb_g, 2016-07-22 19:36:13]

Fix win32 bug.

[herb_g, 2016-07-22 19:37:47]

The CQ bit was checked by herb@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-07-22 19:37:55]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-22 19:37:58]

Your CL relies on deprecated CQ feature(s):
 * Specifying master names in CQ_INCLUDE_TRYBOTS part of description without
"master." prefix is deprecated:
     client.skia
   For more details, see http://crbug.com/617627.

[commit-bot: I haz the power, 2016-07-22 20:05:22]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-07-22 20:05:23]

Dry run: This issue passed the CQ dry run.

[herb_g, 2016-07-22 20:35:42]

Document bug.

[herb_g, 2016-07-22 20:36:27]

The CQ bit was checked by herb@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-07-22 20:36:38]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-22 20:36:41]

Your CL relies on deprecated CQ feature(s):
 * Specifying master names in CQ_INCLUDE_TRYBOTS part of description without
"master." prefix is deprecated:
     client.skia
   For more details, see http://crbug.com/617627.

[herb_g, 2016-07-22 20:37:22]

PTAL to check in for the MSVC 2015u2 bug.

[mtklein, 2016-07-22 20:44:54]

lgtm

[commit-bot: I haz the power, 2016-07-22 20:59:11]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-07-22 20:59:12]

Dry run: This issue passed the CQ dry run.

[herb_g, 2016-07-22 21:05:22]

The CQ bit was checked by herb@google.com

[commit-bot: I haz the power, 2016-07-22 21:05:34]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-07-22 21:05:36]

Your CL relies on deprecated CQ feature(s):
 * Specifying master names in CQ_INCLUDE_TRYBOTS part of description without
"master." prefix is deprecated:
     client.skia
   For more details, see http://crbug.com/617627.

[herb_g, 2016-07-22 21:05:50]

Description was changed from

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed:
https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
==========

to

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:5566
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed:
https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
==========

[commit-bot: I haz the power, 2016-07-22 21:06:30]

Description was changed from

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:5566
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed:
https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
==========

to

==========
In the current code, tiling and bilerp sampling are strongly tied together. They
can be separated by taking advantage of observation that translating a sample
point into filter points in the bilerp stage the filter points will be at most
0.5 outside the tile. This allows simplified repositioning for the various
tiling modes; clamp and mirror use min and max while repeat has max -> 0 and 0->
max. This allows bilerp to simply treat the filter points that fall off the
tile. This allows tiling and bilerp sampling to be totally separate.

This CL has several parts that are intertwined:
* move pin/wrap functionality into BilerpSampler.
* remove the nearest neighbor and bilerp tilers
* create a simplified general tiler
* remove the pipeline virtual calls bilerpEdge and bilerpSpan because everything
works of sample points now.
* redo all the bilerp sampling to use the new local to methods to wrap/pin.
* introduce a new medium rate sample that handles spans with 1 < |dx| < 2.

This change improves the performance as displayed below:
Most of top 25 desktop improves or are the same. A few are worse, but close to
the noise floor. In addition, this change has about 3% smaller code.

old time     new time   new/old
13274693  8414645  0.633886  top25desk_google_com_search_q_c.skp_1
4946466   3258018  0.658656  top25desk_wordpress.skp_1
6977187   5737584  0.822335  top25desk_youtube_com.skp_1
3770021   3296831  0.874486  top25desk_google_com__hl_en_q_b.skp_1
8890813   8600143  0.967307  top25desk_answers_yahoo_com.skp_1
3178974   3094300  0.973364  top25desk_facebook.skp_1
8871835   8711260  0.981901  top25desk_twitter.skp_1
838509    829290   0.989005  top25desk_blogger.skp_1
2821870   2801111  0.992644  top25desk_plus_google_com_11003.skp_1
511978    509530   0.995219  top25desk_techcrunch_com.skp_1
2408588   2397435  0.995369  top25desk_ebay_com.skp_1
4446919   4448004  1.00024   top25desk_espn.skp_1
2863241   2875696  1.00435   top25desk_google_com_calendar_.skp_1
7170086   7208447  1.00535   top25desk_booking_com.skp_1
7356109   7417776  1.00838   top25desk_pinterest.skp_1
5265591   5340392  1.01421   top25desk_weather_com.skp_1
5675244   5774144  1.01743   top25desk_sports_yahoo_com_.skp_1
1048531   1067663  1.01825   top25desk_games_yahoo_com.skp_1
2075501   2115131  1.01909   top25desk_amazon_com.skp_1
4262170   4370441  1.0254    top25desk_news_yahoo_com.skp_1
3789319   3897996  1.02868   top25desk_docs___1_open_documen.skp_1
919336    949979   1.03333   top25desk_wikipedia__1_tab_.skp_1
4274454   4489369  1.05028   top25desk_mail_google_com_mail_.skp_1
4149326   4376556  1.05476   top25desk_linkedin.skp_1

BUG=skia:5566
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2134893002
CQ_INCLUDE_TRYBOTS=client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot;master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot

Committed:
https://skia.googlesource.com/skia/+/8602ede5fdfa721dcad4dcb11db028c1c24265f1
Committed:
https://skia.googlesource.com/skia/+/86a6c6d8bdb1055ee2d0689eeaa4a814402e69dd
==========

[commit-bot: I haz the power, 2016-07-22 21:06:31]

Committed patchset #17 (id:320001) as
https://skia.googlesource.com/skia/+/86a6c6d8bdb1055ee2d0689eeaa4a814402e69dd